RESEARCH ARTICLE

Effect of a community-based approach of iron

and folic acid supplementation on compliance

by pregnant women in Kiambu County,

Kenya: A quasi-experimental study

Mary Wanjira KamauID1*, Samuel Thuo Kimani1, Waithira Mirie1, Isaac Kamau Mugoya2

1 School of Nursing Sciences, University of Nairobi, Nairobi, Kenya, 2 John Snow Inc., Nairobi, Kenya

* kwanjira@uonbi.ac.ke

Abstract

Introduction

Iron and Folic Acid Supplementation (IFAS) is an essential and affordable intervention strat-

egy for prevention of anaemia during pregnancy. The supplements are currently provided

for free to pregnant women in Kenya during antenatal care (ANC), but compliance remains

low over the years. There is need for diversification of IFAS programme implementation by

exploring other distribution channels to complement existing antenatal distribution and

ensure consistent access to IFAS supplements.

Objectives

To determine the effect of a community-based approach of IFAS distribution on compliance

and assess side-effects experienced and their mitigation by pregnant women in Kiambu

County.

Methodology

A pretest-posttest quasi-experimental study design was used, consisting of an intervention

and a control group, among 340 pregnant women 15–49 years, in five health facilities in Lari

Sub-County in Kiambu County, between June 2016 and March 2017. Community health vol-

unteers provided IFAS supplements, counselling and weekly follow-up to pregnant women

in the intervention group while the control group followed standard practice from health facili-

ties. Baseline and endline data were collected during antenatal care and compared. Quanti-

tative data was analyzed using STATA version 14. Analysis of effect of intervention was

done using Difference-In-Difference regression approach.

Results

Levels of compliance increased by 8% in intervention group and 6% in control group. There

was increased awareness of IFAS side-effects across groups. The intervention group

PLOS ONE | https://doi.org/10.1371/journal.pone.0227351 January 10, 2020 1 / 17

a1111111111

a1111111111

a1111111111

a1111111111

a1111111111

OPEN ACCESS

Citation: Kamau MW, Kimani ST, Mirie W, Mugoya

IK (2020) Effect of a community-based approach

of iron and folic acid supplementation on

compliance by pregnant women in Kiambu County,

Kenya: A quasi-experimental study. PLoS ONE 15

(1): e0227351. https://doi.org/10.1371/journal.

pone.0227351

Editor: Bishwajit Ghose, University of Dhaka,

BANGLADESH

Received: April 27, 2019

Accepted: December 17, 2019

Published: January 10, 2020

Copyright: © 2020 Kamau et al. This is an open

access article distributed under the terms of the

Creative Commons Attribution License, which

permits unrestricted use, distribution, and

reproduction in any medium, provided the original

author and source are credited.

Data Availability Statement: The data underlying

this study is available from Open Science

Framework: Community based IFAS, Combined

data_PLOS. DOI 10.17605/osf.io/x8tj3. This

dataset is available under a CCO 1.0 Universal

license.

Funding: This research was supported by the

Consortium for Advanced Research Training in

Africa (CARTA). CARTA is jointly led by the African

Population and Health Research Center and the

reported experiencing less side-effects and were better able to manage them compared to

the control group.

Conclusion

Implementation a community-based approach improved maternal compliance with IFAS,

awareness of IFAS side effects and their management, with better improvement being

recorded in the intervention group. Hence, there is need to integrate community-based

approach with antenatal distribution of IFAS to improve supplementation.

1. Introduction

Iron and folic acid are essential micronutrients whose deficiency causes anaemia [1] especially

during pregnancy when the demand for these nutrients is higher than usual. With the global

prevalence of between 41.8% and 43.8% [2], anaemia during pregnancy is a major contributor

to the global burden of disease. The prevalence of anaemia is highest in Africa at 61.3% fol-

lowed by 52.5% in South East Asia [3]. In developing countries, it is estimated that every sec-

ond pregnant woman is anaemic [4]. In Kenya, anaemia in pregnancy remains a public health

problem with 55.1% of the pregnant women being anaemic [5], and iron deficiency accounting

for approximately 60% [6].

If not addressed, anaemia, can eventually cause maternal, foetal and infant death. Globally,

1 out of 5 maternal deaths are associated with anaemia [7]. In Kenya, anaemia in pregnancy is

associated with estimated 1 in 5 perinatal deaths and 1 in 10 maternal deaths respectively [8].

A great proportion of these deaths can be averted by strengthening Iron and Folic Acid Sup-

plementation (IFAS) and increasing its compliance by pregnant women. Iron and folic acid

supplementation has been shown to be beneficial by reducing maternal and infant morbidity

(including postpartum haemorrhage and poor birth outcomes, such as preterm births and low

birth weights) and mortality associated with anaemia [1, 9].

Iron and folic acid supplementation increases haemoglobin levels thereby reducing anae-

mia rates and improving birth outcomes [1, 10]. Research shows that it reduces risk of mater-

nal anaemia and iron deficiency at term by 70% and 57% respectively [11]. In addition,

supplementation with Iron and Folic Acid (IFA) tablets can increase the mean blood haemo-

globin level by 10�2 g/l in pregnant women and by 8�6 g/l in non-pregnant, consequently elimi-

nating about 50% of anaemia in women [12, 13] as well as anaemia risk among infants [14].

Furthermore, intake of 90–150 IFA tablets can decrease neonatal mortality by 34%-45% [15–

17]. Maternal and child mortality are two most important indicators of health, which if

improved, would go a long way in achieving Universal Health Coverage (UHC) against anae-

mia, in line with the mission of UHC 2030 to accelerate both equitable and sustainable prog-

ress towards UHC globally. Consequently, this will contribute to achievement of United

Nations Sustainable Development Goal 3: Good health and well-being for all.

The current mode of distribution of IFA tablets through health facilities only has not been

able to attain the required coverage. There is therefore need to diversify IFAS distribution

channels. Following the World Health Organization (WHO) recommendations, Kenya

adopted oral iron and folic acid supplementation programme in 2010, as a high impact nutri-

tion intervention to specifically control anaemia in pregnancy [18]. Kenya’s efforts to improve

uptake of IFAS have included adoption of a combined iron and folic acid tablet in 2012 and

provision of the supplements free of charge to pregnant women at all public health facilities.

Community-based approach of iron and folic acid supplementation

PLOS ONE | https://doi.org/10.1371/journal.pone.0227351 January 10, 2020 2 / 17

University of the Witwatersrand and funded by the

Carnegie Corporation of New York (Grant No-B

8606.R02), Sida (Grant No:54100029), the

DELTAS Africa Initiative (Grant No: 107768/Z/15/

Z). The DELTAS Africa Initiative is an independent

funding scheme of the African Academy of

Sciences (AAS)’s Alliance for Accelerating

Excellence in Science in Africa (AESA) and

supported by the New Partnership for Africa’s

Development Planning and Coordinating Agency

(NEPAD Agency) with funding from the Wellcome

Trust (UK) (Grant No:107768/Z/15/Z) and the UK

government. The statements made and views

expressed are solely the responsibility of the fellow.

The funders had no role in study design, data

collection and analysis, decision to publish, or

preparation of the manuscript and only provided

financial support in form of research materials to

carry out the study.

Competing interests: IM is affiliated with John

Snow, Inc. This does not alter our adherence to all

the PLOS ONE policies on sharing data and

materials.

However, compliance with IFAS has remained low over the years and prevalence of anaemia

in pregnancy remains persistently high (55.1%) [5]. The 2014 Kenya Demographic Health Sur-

vey indicated that less than 8% pregnant women took IFA supplements for 90 or more days

and over 30% did not take them at all [19]. Community based distribution of IFA tablets cou-

pled with closer follow-up through home visits, if adopted, can play a role in improving com-

pliance and eventually reducing the burden of anaemia in pregnancy.

Studies have shown that side effects to daily IFA tablets are a major reason for low compli-

ance and a barrier to full realization of benefits of supplementation [3, 20–23]. The most men-

tioned sides effects include gastrointestinal (such as epigastric pain, nausea, vomiting,

diarrhea, constipation or gastritis) and dark stools [5, 18, 24]. However, most studies done do

indicate the IFAS side-effects that pregnant women experience but not how they mitigate or

handle them. Clients concerns are better addressed during issuance of the tablets. However,

many health care providers do not even inform clients that they may experience side effects

and clients become non-compliant with any slight side effect experienced [5]. A previous

study indicated that the severity and frequency of occurrence of side effects increases with the

amount/dose of iron administered [25], although this variation was not observed when the

dose was less than 100mg. Consequently, the combined IFA tablet has less amount of iron per

tablet (60mg) compared with iron only tablet (200mg). This when combined with folic acid

into one tablet reduces both side effects experienced by pregnant women and pill burden. This

was expected to increase compliance by pregnant women, but as indicated earlier, the compli-

ance remains low nationally.

Previous studies recommended adoption of community-based delivery of IFAS [26] and its

integration into the formal health-care system [25] because of its contribution in improving

compliance. Experiences in Nicaragua, Gambia, Indonesia and Thailand, have proved that

community based IFAS distribution can help achieve higher rates of compliance by reaching

more women than antenatal (ANC) distribution alone [27]. Further evidence from Ghana and

Malawi indicate that community involvement in anaemia control interventions is critical to

their success [28, 29]. Community based interventions involve use of volunteers/community

agents, as point of contact/intermediary between intervention/health service and clients at

community level. Indeed, since the beginning of the 21st century, many health programs

started utilizing Community Health Workers (CHWs) because of their unique capacity to con-

nect community members with appropriate health care services. The CHWs serve as "bridges"

between the health care services and community members. The CHWs can develop partner-

ships with formal health care delivery systems and work with health care providers to support

community participation in health activities for the success of health programmmes.

In developing countries, two factors that have been identified to substantially contribute to

low compliance with IFAS during pregnancy are poor awareness about IFAS and anaemia as

well as insufficient health delivery systems [3]. In Kenya, IFAS is currently provided during

antenatal care clinic visits only. Adequacy of antenatal care is a major determinant of IFAS uti-

lization during pregnancy [30]. Studies have shown that low access and poor utilization of

antenatal care (ANC) services is highly associated with low utilization of IFAS [31, 32]. Most

pregnant women in Kenya start ANC services late [8] and therefore do not benefit optimally

from the IFAS. This points to the need to diversify delivery strategies beyond ANC attendance

in order to reach more women. The best health outcomes have been realized when the supple-

mentation uses community-based distribution systems [33]. Use of CHWs for IFAS distribu-

tion is one such option.

Research has shown that CHWs are able to identify pregnant women early in pregnancy

and can therefore provide supplements to these women early in pregnancy when they are

most beneficial [7, 9]. Community health workers are in constant touch with the women and

Community-based approach of iron and folic acid supplementation

PLOS ONE | https://doi.org/10.1371/journal.pone.0227351 January 10, 2020 3 / 17

can therefore provide regular and consistent follow up and improve compliance. This

approach has not been previously used in Kenya. There remains a gap in integration of com-

munity based with formal IFAS distribution strategies in Kenya, which formed the basis for

this study. In an effort to bridge this gap and address the insufficient IFAS delivery channel,

this study aimed at introducing a community based IFAS distribution strategy to complement

the existing health facility based IFAS distribution in order to increase its utilization by preg-

nant women in the community. This study used CHWs, otherwise referred to as Community

Health Volunteers (CHVs) in Kenya to distribute IFAS tablets to the pregnant women in their

homes and to counsel them. Therefore, the objectives of this study were to (1) determine the

effect of a community based IFAS distribution on compliance (2) identify IFAS side-effects

experienced by pregnant women and (3) assess mitigation measures of IFAS side-effects by

pregnant women, before and after community-based distribution of IFAS.

2. Materials and methods

2.1 Study site

This study was conducted between June 2016 and March 2017, in Kenya, Kiambu County,

Lari Sub-County, in five of its major public health facilities (Lari, Githirioini, Kagwe, Kagaa

and Kinale).

2.2 Study design and structure

This was a pretest-posttest quasi-experimental study design with a control group. The study

involved three phases as shown in Fig 1, namely inception, implementation and follow-up

phases. Inception was the first phase and involved identification and recruitment of study

respondents as well as baseline data collection including: socio-demographic characteristics,

compliance with IFAS, side-effects experienced with IFAS and their mitigation measures by

pregnant women. Implementation was the second phase and involved training of Health Care

Providers (HCPs), mostly nurses and Community Health Volunteers (CHVs) on IFAS pro-

gramme then distribution of IFAS supplements with counselling information. The control

group followed standard routine practice of receiving IFAS tablets from HCPs during antena-

tal care clinics. The intervention group received IFAS tablets from CHVs who distributed the

IFAS tablets to pregnant women in their homes. Follow-up was the third phase and involved

following the pregnant women up to delivery of their babies as well as collection of endline

data. The control group were followed up by HCPs (specifically nurses) during routine antena-

tal care in health facilities. The intervention group were followed up by CHVs on a weekly

basis in their homes. During the weekly home visits, CHVs provided each pregnant woman

with the entire week’s supply of IFAS tablets, evaluated previous week’s intake and counselled

her on various IFAS topics including common side effects such as black stools, stomach upset,

constipation and diarrhea, and their mitigation measures. In addition, the CHVs encouraged

pregnant women to attend antenatal care clinics to receive the other antenatal care services.

Using a similar questionnaire like that used to collect baseline data, endline data was then col-

lected before delivery, from 36th week of gestation, in both control and intervention groups.

2.3 Sampling and study population

Some of the details of methods adopted for recruitment of the pregnant women involved in

this study as well as the ethical considerations have been published elsewhere in other articles

[34, 35].

Community-based approach of iron and folic acid supplementation

PLOS ONE | https://doi.org/10.1371/journal.pone.0227351 January 10, 2020 4 / 17

From Kiambu County, two stage sampling method was used to select one Sub-County

(Lari) and five of its major public health facilities (Lari, Githirioini, Kagwe, Kagaa and Kinale).

The sampling frame consisted of all Sub-Counties in Kiambu County. Lari Sub-County was

selected on the basis of having existing functional (active) community units, meaning its com-

munity health volunteers were actively involved in provision of community health services to

community members. The five health facilities were selected on the basis of existing functional

(active) community units attached to them and high client population turnover, due to the low

turnover of ANC clients.

The study sample size was calculated using the following formula for a binary outcome

[36]:

n ¼ ðr þ 1

rÞð�pÞð1 � �pÞðZb þ Za=2Þ

2

ðDÞ2

D is the expected effect in IFAS compliance of 20% (control 25% to 45% intervention)

Fig 1. Study phases. This figure shows the three stages that were followed in the implementation of this study.

https://doi.org/10.1371/journal.pone.0227351.g001

Community-based approach of iron and folic acid supplementation

PLOS ONE | https://doi.org/10.1371/journal.pone.0227351 January 10, 2020 5 / 17

A consideration of 30% loss to follow-up was added to this sample, making a total sample

size of 170. The final sample size per study group was 170 and in both groups was therefore

340.

The study population consisted of all pregnant women who attended antenatal care clinic

in the five health facilities. The inclusion criteria was: age 18–49 years, below 33 weeks in their

pregnancy gestation, not suffering from any chronic illness and who provided informed con-

sent to participate in the study. Consecutive sampling method, which is considered the best

type of non-probability sampling with best representation of entire population, was used to

include all accessible pregnant women as part of the sample. All pregnant women who met the

inclusion criteria were informed about the study and those who provided both verbal and writ-

ten informed consent to participate in the study were recruited. Those residing in a commu-

nity that had a functional community unit with active community health volunteers who

consented to have IFAS distributed to them in their homes formed the intervention group.

Those who consented to participate and were residing in a community that did not have a

functional community unit, formed the control group, who received their IFAS from fixed

health facilities during antenatal care clinics, until the required sample size of 170 was reached.

2.4 Data collection tools and process

A semi-structured interviewer-administered questionnaire consisting of 23 closed ended ques-

tions including; 11 on socio-demographic data, and 12 on IFAS utilization at health facility,

was developed, pre-tested and used for data collection in this study. To address any potential

bias in data collection, four research assistants were trained on research ethics, protocols and

quality data collection at Kiambu level 5 hospital where the research questionnaires were

pretested.

To ensure reliability of the questionnaire, a test re-test method was adopted in pre-testing,

whereby a repeat pre-test was conducted after two weeks, and Cohen’s kappa statistic was used

to measure the level of agreement of the results from the two pre-tests. The questions which

were re-tested included: on socio-demographic data: age, education level, occupation, income,

gestation, parity and gravidity; on IFAS utilization: frequency of taking IFAS, duration of tak-

ing IFAS, timing when IFAS was taken, number of tablets taken in the past 7 days, IFAS side

effects experienced and how the pregnant women mitigated the side effects. All the questions

repeated had a kappa value of above 0.7 after comparison thus the questionnaire was consid-

ered reliable, hence all the questions were retained. To ensure validity of the questionnaire, it

was shared and discussed with experts from the Ministry of Health, division of nutrition, and

the study supervisors. The feedback obtained from these experts and pre-testing results was

used to refine the tool and improve its quality to ensure the questions were able to test what

was intended.

The trained research assistants then administered questionnaires to all pregnant women

who met the inclusion criteria and consented to participate in the study.

2.5 Determination of respondents’ compliance levels

The compliance with IFAS was assessed based on the reported number of IFAS tablets taken

in the preceding one week (seven days) before the interview. The IFAS compliance status was

defined as the number of IFAS tablets taken in the preceding seven (7) days. Pregnant women

who took at least 70% of the expected dose of the IFAS tablets in the week preceding the inter-

view, an equivalent of five IFAS tablets per week, were considered compliant with IFAS [37,

38]. Conversely, the respondents who took less than 30% of the expected IFAS dose, an equiva-

lent of less than five IFAS tablets, were considered non-compliant. Women who had attended

Community-based approach of iron and folic acid supplementation

PLOS ONE | https://doi.org/10.1371/journal.pone.0227351 January 10, 2020 6 / 17

the first ANC visit in the current pregnancy and reported not to have received IFAS tablets

before, were excluded in the assessment of compliance since pregnant women routinely

receive IFAS during ANC visits.

2.6 Data management and analysis

In order to examine effectiveness of the intervention, baseline and endline surveys were con-

ducted in both study groups, using a similar questionnaire. To ensure adherence to optimal

data quality standards, the researcher closely supervised the research assistants.

Quantitative data at both baseline and endline was coded after collection then entered into

the computer, cleaned and validated using Statistical Package for Social Sciences (SPSS) statis-

tical software version 22. Data entry was done during the study data collection process to mini-

mize errors. The entered data was then exported to STATA version 14 for analysis. To ensure

confidentiality, the computer access was restricted by password protection. Each questionnaire

had a unique identifier to allow validation. Data cleaning and validation was done prior to

analysis.

Descriptive statistics, including univariate analysis: simple proportions, n (%), for categori-

cal variables and mean with standard deviation for continuous variables, were reported at

baseline and endline. Characteristics of respondents were also described in both intervention

and control groups. To ensure the change caused by the intervention was not by chance, simi-

larity of baseline characteristics in both groups was ascertained. Homogeneity of study groups

at baseline was determined by comparing socio-demographic characteristics of both groups.

Side-effects experiences and the measures applied by pregnant women to mitigate them were

recorded. Bivariate analysis, using the chi-square test, was done for comparison between

groups and multivariate analysis was used to control for confounders.

The analysis of effect of the intervention was done using a Difference-In-Difference (DID)

regression model to compare outcomes between intervention and control groups before (base-

line) and after (endline) the intervention. The changes in the dependent variables in the inter-

vention group (from baseline to endline) were compared to changes in the control group

(from baseline to endline) as shown in Table 1 below [39]. The intervention effect was mea-

sured by odds ratio and 95% confidence level of the interaction term between study groups

(intervention and control) and period of survey (baseline and endline) in the regression

model. A p-value of 0.05 was considered statistically significant. Since the same respondents

who participated in the baseline are the same who participated in the end term evaluation, the

analysis adopted a paired analysis with repeated measures instead of treating the respondents

in baseline and endline as independent groups.

2.7 Ethical considerations

Scientific and ethical approval was sought and obtained from Kenyatta National hospital/Uni-

versity of Nairobi Ethics and Research Committee (KNH-ERC/A/90 protocol number–P706/

11/2015). Research permit was sought and obtained from the National Commission for Sci-

ence, Technology and Innovation (NACOSTI/P/18/81499/22319). Authority to conduct the

study was obtained from Kiambu County, Lari Sub-county authorities and all health facilities

involved.

Respondents were fully protected from any form of harm. Participation in the study was

purely voluntary. The purpose of study was clearly explained to respondents who were

required to provide informed verbal and written consent. Emphasis on confidentiality and pri-

vacy were made clear at the time of consenting to participate and upheld throughout the study.

No name appeared on the questionnaires so no participant identification with information

Community-based approach of iron and folic acid supplementation

PLOS ONE | https://doi.org/10.1371/journal.pone.0227351 January 10, 2020 7 / 17

could occur. Respondents were at liberty to discontinue from the study at any time without

facing any adverse consequences. Information was kept confidential by restricted access and

coding of questionnaires.

3. Results

3.1 Socio-demographic characteristics of study respondents

A total of 340 pregnant women participated in the study during the baseline, of these 189

(56%) participated during the endline. Table 2 shows the socio-demographic characteristics of

study respondents at baseline according to study group. Most (n = 212, 62%) respondents

were 20–29 years of age, with mean age of 25.6 (SD ± 5.6), had a secondary level of education

(n = 180, 53%) and unemployed (n = 167, 49%). Majority of them were married (n = 288,

85%) and earned less than USD. 100 per month (n = 305, 93%). Whereas only 6% (n = 20) had

attained tertiary level of education, only 3% (n = 10) were formally employed. In terms of gra-

vidity, most (n = 223, 68%) of the women were multigravida. The study groups were homoge-

neous at baseline (Table 2) since there was no statistical difference (p>0.05) in the baseline

characteristics of respondents between the two comparison groups.

3.2 Effect of community-based approach on maternal compliance with

IFAS

3.2.1 Compliance with IFAS among pregnant women. In this study, respondents who

took at least 70% (5 tablets) of the expected dose of IFAS tablets in the week preceding the

interview were considered compliant with IFAS as earlier explained. Fig 2 shows the levels of

compliance with IFAS between baseline and endline across the two groups. There was an

improvement in compliance with IFAS in both groups at endline. Levels of compliance

increased by 8 percentage points (from 63.8% to 71.4%) and 6 percentage points (from 68.5%

to 74.3%) in the intervention and control group, respectively. The intervention had a net effect

of 2 percentage points (8–6) increase in compliance. However, it did not yield statistical differ-

ence since the DID between the two groups was 0.02 and the CI was -0.20, 0.24.

3.2.2 Factors associated with maternal compliance with IFAS. Table 3 shows that the

logistic regression to assess the effect of the intervention and other potential factors on mater-

nal compliance with IFAS did not show statistical significance at p-value of 0.05 in compliance

with IFAS between the intervention and control groups.

3.3 Side effects experienced with IFAS and their mitigation by respondents

The results in Fig 3 shows a comparison between baseline and endline proportions of side-

effects experienced by each study group. There was increased awareness of the specific side

effects of IFAS in both groups during the study period and respondents were better able to

identify the side-effects associated with IFAS at endline than at baseline. For example, at base-

line, none associated faeces turning black with IFAS, which is normal as a result of iron

Table 1. Intervention effect formula.

Baseline Endline

Intervention group Level of phenomenon before intervention (X) Intervention introduced Level of phenomenon after intervention (Y)

Control group Level of phenomenon without intervention (A) Level of phenomenon without intervention (Z)

Intervention Effect = (Y-X)–(Z-A)

Source: Kothari and Garg, 2014 pg 41

https://doi.org/10.1371/journal.pone.0227351.t001

Community-based approach of iron and folic acid supplementation

PLOS ONE | https://doi.org/10.1371/journal.pone.0227351 January 10, 2020 8 / 17

absorption and harmless. But at endline, they were able to identify it. In addition, the number

of side effects reported was generally much lower in the intervention group compared to the

control group at endline, unlike at baseline where the intervention group reported more side

effects. Similarly, there was greater awareness on the mitigation of IFAS side effects in the

intervention group than in the control group, as shown in Fig 4. Notably, in the intervention

group, there was a greater decrease in the proportion of respondents who stopped taking IFAS

(22.6%-9.8%) on experiencing side-effects. Moreover, there was greater change in the practices

of managing IFAS side effects such as taking IFAS with meals (6.5%-15.7%) and taking IFAS at

bedtime (3.2%-17.7%), in the intervention group compared to the control group.

4. Discussion

The aim of this study was to determine the effect of a community based IFAS distribution

strategy on compliance with IFAS and identify side-effects experienced before and after study

as well as the measures pregnant women used to mitigate these side-effects. The findings indi-

cate that (1) there was a better improvement in compliance in the intervention group (2) there

Table 2. Socio-demographic profile of the study respondents at baseline by group.

Variable Total (Col %) N = 340 Hospital (Row %) N = 218 Community (Row %) N = 122 Chi-square p-value

Age of pregnant woman in years

Less than 20 years 43 (12.6) 27 (62.8) 16 (37.2) 0.899

20–29 years 212 (62.4) 138 (65.1) 74 (34.9)

30 years and above 85 (25) 53 (62.4) 32 (37.6)

Mean age (std) 25.6 (5.6) 25.6 (5.9) 25.7 (5.7)

Marital status

Married 288 (84.7) 183 (63.5) 105 (36.5) 0.32

Single 51 (15) 35 (68.6) 16 (31.4)

Education level

Primary 137 (40.7) 89 (65) 48 (35) 0.467

Secondary 180 (53.4) 111 (61.7) 69 (38.3)

Tertiary 20 (5.9) 15 (75) 5 (25)

Occupation of pregnant woman

Unemployed 167 (49.1) 108 (64.7) 59 (35.3) 0.978

Casual employment 77 (22.6) 49 (63.6) 28 (36.4)

Self-employed/Employed 96 (28.2) 61 (63.5) 35 (36.5)

Average income per month in USD

Less than100 305 (93) 194 (63.6) 111 (36.4) 0.274

100 and above 23 (7) 12 (52.2) 11 (47.8)

Parity

0 111 (33) 74 (66.7) 37 (33.3) 0.106

1 92 (27.4) 53 (57.6) 39 (42.4)

2 80 (23.8) 50 (62.5) 30 (37.5)

3 and above 53 (15.8) 41 (77.4) 12 (22.6)

Gravidity

Primigravida 107 (32.4) 74 (69.2) 33 (30.8) 0.171

Multigravida 223 (67.6) 137 (61.4) 86 (38.6)

Religion of pregnant woman

Protestant Christian 283 (83.5) 179 (63.3) 104 (36.7) 0.512

Catholic Christian 56 (16.5) 38 (67.9) 18 (32.1)

https://doi.org/10.1371/journal.pone.0227351.t002

Community-based approach of iron and folic acid supplementation

PLOS ONE | https://doi.org/10.1371/journal.pone.0227351 January 10, 2020 9 / 17

was increased awareness on IFAS side effects across the study groups (3) the intervention

group reported fewer side-effects than the control group at endline (4) the intervention group

reported better mitigation of side-effects than the control group at endline. These findings are

consistent with literature on strengths of community based distribution of IFAS as a valuable

platform in implementation of IFAS programmes [7].

Compliance with IFAS improved during the study period, indicating increase in IFAS

uptake, with the intervention group showing higher improvement than the control group.

This means there was an improvement in the proportion of women who consistently took the

IFAS tablets as recommended. These findings were similar to a coverage of 67% reported in

urban Nairobi County [40] in Kenya and 64.7% reported in India [41] respectively. However,

the findings were higher than reports on IFAS coverage from low and medium income coun-

tries (LMIC) with similar rural settings namely, South Ethiopia (39.2%), Pakistan (38.3%),

[38], Nigeria (37.5%) [42] and Western Ethiopia (20.4%) [20]. Nevertheless, lower coverage

(18%) has been reported in a neighbouring County of Machakos, Kenya [43]. Elsewhere, Bili-

male and colleagues suggested adoption of “more open, cooperative health professional-

Fig 2. Compliance with IFAS among pregnant women. This refers to the level of maternal compliance with IFAS at both baseline and endline for both the control

(hospital) and intervention (community) groups.

https://doi.org/10.1371/journal.pone.0227351.g002

Community-based approach of iron and folic acid supplementation

PLOS ONE | https://doi.org/10.1371/journal.pone.0227351 January 10, 2020 10 / 17

patient relationships” as critical to improving compliance [3]. Addressing all clients’ needs and

issues through open communication is highly recommended. This has been achieved in this

study by using CHVs during weekly IFAS distribution and follow ups of pregnant women at

home. Since both HCPs and CHVs were trained in this study to provide IFAS tablets and

counsel, both intervention and control groups yielded similar improvement in compliance lev-

els although it was slightly higher in the intervention group. This highlights the importance of

community-based distribution of IFAS and the need to adopt the community-based approach.

This study showed increased awareness on IFAS side effects in both groups. Studies have

shown that accompanying health education with clear instructions on intake of IFAS substan-

tially improves compliance [38, 44]. This is demonstrated in another Kenyan study where

advice by health workers was independently associated with higher compliance [37, 45]. This

implies that counselling on IFAS contributed to improving compliance among pregnant

women in this study. Quality counselling should therefore accompany distribution of

supplements.

Findings from this study showed a decrease in side effects reported at endline compared to

baseline, especially in the intervention group. In addition, the study showed more awareness

of mitigation of IFAS side effects in the intervention group. Studies have associated awareness

of side effects of IFAS and their management with higher compliance [3, 20–22]. This is

because side effects experienced on taking IFAS are associated with poor compliance. Many

HCPs fail to inform clients about side-effects of IFAS resulting in either poor or non-compli-

ance when they experience any slight discomfort [5]. Educating pregnant women on how to

mitigate the side effects of IFAS often leads to higher compliance [3, 21, 46]. Many pregnant

women stop taking IFAS tablets when they experience side effects. It is therefore critical to

Table 3. Factors associated with maternal compliance with IFAS.

Variable Odds Ratio P-value 95% Confidence Interval

Community vs. hospital 0.936 0.873 0.416 2.106

Endline vs. Baseline 1.634 0.172 0.807 3.310

Interaction (group, time)�� 0.764 0.635 0.251 2.321

Age group

20–29 years vs. <20 years 1.121 0.796 0.471 2.665

� 30 years vs. <20 years 1.133 0.833 0.354 3.631

Highest education level

Secondary vs. Primary 0.900 0.736 0.489 1.658

Tertiary vs. Primary 0.680 0.550 0.191 2.413

Occupation

Casual vs. Unemployed 1.289 0.526 0.588 2.827

Employed vs. Unemployed 0.759 0.402 0.399 1.445

Single vs. Married 1.528 0.388 0.584 4.000

Income in USD

�100 vs. <100 1.524 0.412 0.557 4.164

Previous pregnancies

1 vs. 0 1.424 0.597 0.384 5.280

2 vs. 0 2.140 0.289 0.525 8.729

3 vs. 0 3.320 0.141 0.671 16.424

Multigravida vs. Primigravida 0.371 0.123 0.105 1.310

Catholic vs. Protestant 1.627 0.267 0.689 3.840

��DID

https://doi.org/10.1371/journal.pone.0227351.t003

Community-based approach of iron and folic acid supplementation

PLOS ONE | https://doi.org/10.1371/journal.pone.0227351 January 10, 2020 11 / 17

make pregnant women aware of the possible side effects and how to mitigate them in order to

improve adherence to IFAS as indicated in other studies [3, 21].

This study showed better mitigation of IFAS side-effects in the intervention group.

Although health workers may generally assume that talking about IFAS side-effects will make

pregnant women not take IFAS, proper counselling and open communication leads to greater

utilization as this study indicates. This was also demonstrated in a study among Indonesian

women, who were not deterred by IFAS side-effects following prior discussion of likely side

effects. This study indicates that side-effects and their mitigation measures is one of the IFAS

topical areas that requires strengthening for quality IFAS counselling. This will eventually

enable pregnant women to mitigate IFAS side-effects more effectively thus reducing their

occurrence and improving their management whenever they occur, rather than discontinuing

IFAS tablets which leads to poor or non-compliance.

This study had several limitations. (1) Following clients for a long period of time meant that

some pregnant women were lost to follow up and this affected the power of the study. This was

minimized by increasing the sample size by an estimated rate of loss to follow-up of 30%.

Despite this consideration, there was a high loss to follow-up from 340 at baseline to 189 at

endline mainly because of the extended industrial actions (strikes) among health workers in

public health facilities during the study period. The industrial actions led to disruption of ser-

vices in public health facilities that led to many clients seeking for antenatal and other health

services from private and/or mission health facilities. (2) The study design did not involve ran-

domization. The intervention was community based, using the natural setting and did not

modify the environment of the respondents. The effect of this limitation was minimized by use

Fig 3. Side effects experienced by pregnant women with IFAS. This refers to the side effects experienced by pregnant women

taking IFAS at both baseline and endline for both the control (hospital) and intervention (community) groups.

https://doi.org/10.1371/journal.pone.0227351.g003

Community-based approach of iron and folic acid supplementation

PLOS ONE | https://doi.org/10.1371/journal.pone.0227351 January 10, 2020 12 / 17

of a control group to avoid extraneous variation resulting both from passage of time and from

non-comparability of the test and control areas. Also, due to the low turn-out of antenatal cli-

ents in the Sub-County, consecutive sampling was used to include all accessible pregnant

women as part of the sample. (3) Training of both HCPs and CHVs, instead of CHVs only,

was also a limitation since it introduced confounding of the effect of the intervention. This was

necessary as the CHVs needed to refer clients for other antenatal services and were being

supervised and supplied with IFAS tablets by the nurses, so the nurses had to be trained as

well. (4) The study results were prone to recall bias and subjectivity because the study greatly

relied on verbal reports from the interviewees. This challenge was mitigated by training the

interviewers as well as double questioning to identify any inconsistencies in the interview

reports. (5) Generalizations of the study findings to other areas with different socio-demo-

graphic characteristics may be difficult since the study was restricted to one Sub-County.

5. Conclusion

Implementation of a community-based approach of IFAS distribution improved the maternal

compliance with IFAS. Fewer side-effects were reported and were better mitigated especially

in the intervention group. Use of CHVs for IFAS distribution is a potential approach for

Fig 4. Mitigation measures of IFAS side effects by pregnant women. This refers to the measures that pregnant women taking IFAS used to

mitigate IFAS side effects whenever they experienced them, at both baseline and endline for both the control (hospital) and intervention

(community) groups.

https://doi.org/10.1371/journal.pone.0227351.g004

Community-based approach of iron and folic acid supplementation

PLOS ONE | https://doi.org/10.1371/journal.pone.0227351 January 10, 2020 13 / 17

diversification of IFAS implementation in Kenya. It can be used to increase both demand and

delivery of IFAS to pregnant women and to improve the low compliance levels. We recom-

mend adoption of the use of CHVs to distribute IFAS supplements and follow up pregnant

women to ensure improved IFAS uptake and compliance. This is a cost-effective strategy

because CHVs are already operational in communities, so pregnant women will trust them.

Findings from this study indicate and recommend that the time for integration of CBA into

existing vertical health facility approach, to complement antenatal IFAS distribution, is now.

This will further increase supplementation coverage and consequently reduce deficiencies of

these crucial micronutrients among women and children, contributing to universal health cov-

erage and achievement of sustainable development goal 3.

Acknowledgments

Our gratitude to all study participants in Lari Sub-County for their time and willingness to

share their experiences. We thank the Ministry of Health Headquarters, County and Sub-

County management teams for allowing the study to continue in their health facilities. We are

grateful for the dedicated performance of the field and data management staff. Our sincere

appreciation to Consortium for Advanced Research Training in Africa (CARTA) for sponsor-

ing this study.

Author Contributions

Conceptualization: Mary Wanjira Kamau, Isaac Kamau Mugoya.

Data curation: Mary Wanjira Kamau, Samuel Thuo Kimani.

Formal analysis: Samuel Thuo Kimani.

Funding acquisition: Mary Wanjira Kamau.

Investigation: Mary Wanjira Kamau, Isaac Kamau Mugoya.

Methodology: Mary Wanjira Kamau, Samuel Thuo Kimani, Waithira Mirie, Isaac Kamau

Mugoya.

Project administration: Mary Wanjira Kamau, Samuel Thuo Kimani, Waithira Mirie, Isaac

Kamau Mugoya.

Resources: Mary Wanjira Kamau, Isaac Kamau Mugoya.

Supervision: Samuel Thuo Kimani, Waithira Mirie.

Validation: Samuel Thuo Kimani, Waithira Mirie, Isaac Kamau Mugoya.

Visualization: Samuel Thuo Kimani, Waithira Mirie, Isaac Kamau Mugoya.

Writing – original draft: Mary Wanjira Kamau, Isaac Kamau Mugoya.

Writing – review & editing: Samuel Thuo Kimani, Waithira Mirie, Isaac Kamau Mugoya.

References1. WHO. Nutritional Anaemias: Tools for Effective Prevention and Control. Geneva: World Health Ogna-

nization; 2017.

2. Mulambah CS, Siamba DN, Ogutu PA, Siteti DI, Wekesa AW. Anaemia in pregnancy: Prevalence and

possible risk factors in Kakamega County, Kenya. Science Journal of Public Health 2014; 2(3):216–22.

3. Bilimale A, Anjum J, N. Sangolli H, D. Mallapur M. Improving Adherence to Oral Iron Supplementation

During Pregnancy. Australasian Medical Journal. 2010; 3:281–90. https://doi.org/10.4066/AMJ.2010.

291

Community-based approach of iron and folic acid supplementation

PLOS ONE | https://doi.org/10.1371/journal.pone.0227351 January 10, 2020 14 / 17

4. WHO. Micronutrients. Geneva: World Health Organization; 2014.

5. MoH. National Iron and Folic Acid Supplementation; Communication Strategy, 2013–2017. In: Division

of Nutrition, editor. Nairobi2013.

6. MoH. National Food and Nutrition Security Policy; Sessional Paper. In: Division of Nutrition, editor.

Nairobi2011.

7. Kavle JA, Landry M. Community-based distribution of iron–folic acid supplementation in low- and mid-

dle-income countries: a review of evidence and programme implications. Public Health Nutrition. 2017;

21(2):346–54. Epub 10/24. https://doi.org/10.1017/S1368980017002828 PMID: 29061205

8. MoH. Accelerating reduction of iron deficiency anaemia among pregnant women in Kenya: Plan of

action 2012–2017. In: Division of Nutrition, editor. Nairobi2012.

9. Nisar YB, Dibley MJ. Earlier Initiation and Use of a Greater Number of Iron-Folic Acid Supplements dur-

ing Pregnancy Prevents Early Neonatal Deaths in Nepal and Pakistan. PLOS ONE. 2014; 9(11):

e112446. https://doi.org/10.1371/journal.pone.0112446 PMID: 25398011

10. Imdad A, Bhutta ZA. Routine iron/folate supplementation during pregnancy: effect on maternal anaemia

and birth outcomes. Paediatr Perinat Epidemiol. 2012; 26 Suppl 1:168–77. Epub 2012/07/07. https://

doi.org/10.1111/j.1365-3016.2012.01312.x PMID: 22742609.

11. WHO. Guideline: Daily iron and folic acid supplementation in pregnant women. Geneva: World Health

Organization 2012.

12. Casey GJ, Jolley D, Phuc TQ, Tinh TT, Tho DH, Montresor A, et al. Long-Term Weekly Iron-Folic Acid

and De-Worming Is Associated with Stabilised Haemoglobin and Increasing Iron Stores in Non-Preg-

nant Women in Vietnam. PLOS ONE. 2011; 5(12):e15691. https://doi.org/10.1371/journal.pone.

0015691 PMID: 21209902

13. Peña-Rosas JP, De-Regil LM, Garcia-Casal MN, Dowswell T. Daily oral iron supplementation during

pregnancy. Cochrane Database of Systematic Reviews. 2015;(7). https://doi.org/10.1002/14651858.

CD004736.pub5 CD004736. PMID: 26198451

14. Tran TD, Fisher J, Hanieh S, Tran T, Simpson JA, Tran H, et al. Antenatal Iron Supplementation Regi-

mens for Pregnant Women in Rural Vietnam and Subsequent Haemoglobin Concentration and Anae-

mia among Their Infants. PLOS ONE. 2015; 10(4):e0125740. https://doi.org/10.1371/journal.pone.

0125740 PMID: 25928545

15. Titaley CR, Dibley MJ, Roberts CL, Hall J, Agho K. Iron and folic acid supplements and reduced early

neonatal deaths in Indonesia. Bull World Health Organ. 2010; 88. https://doi.org/10.2471/blt.09.065813

PMID: 20616969

16. Titaley CR, Dibley MJ, Roberts CL, Agho K. Combined iron / folic acid supplements and malaria prophy-

laxis reduce neonatal mortality in 19 sub-Saharan African countries. American Journal of Clinical Nutri-

tion. 2010; 92(1):235–43. https://doi.org/10.3945/ajcn.2009.29093 PMID: 20504976.

17. Nisar YB, Dibley MJ. Antenatal iron-folic acid supplementation reduces risk of low birthweight in Paki-

stan: secondary analysis of Demographic and Health Survey 2006–2007. Matern Child Nutr. 2014.

Epub 2014/11/26. https://doi.org/10.1111/mcn.12156 PMID: 25422133.

18. MoH. National Policy Guideline on combined iron and folic acid (IFA) supplementation for pregnant

mothers in Kenya. In: Division of Nutrition, editor. Nairobi2013.

19. KNBS, Macro ICF. Kenya Demographic and Health Survey 2014. Nairobi: Kenya National Bureau of

Statistics; 2015.

20. Taye B, Abeje G, Mekonen A. Factors associated with compliance of prenatal iron folate supplementa-

tion among women in Mecha district, Western Amhara: a cross-sectional study. Pan Afr Med J. 2015;

20:43. Epub 2015/06/20. https://doi.org/10.11604/pamj.2015.20.43.4894 PMID: 26090001; PubMed

Central PMCID: PMC4449983.

21. Gebremedhin S, Samuel A, Mamo G, Moges T, Assefa T. Coverage, compliance and factors associ-

ated with utilization of iron supplementation during pregnancy in eight rural districts of Ethiopia: a cross-

sectional study. BMC public health. 2014; 14:607. https://doi.org/10.1186/1471-2458-14-607 PMID:

24930036; PubMed Central PMCID: PMC4073172.

22. Zavaleta N, Caulfield LE, Figueroa A, Chen P. Patterns of compliance with prenatal iron supplementa-

tion among Peruvian women. Matern Child Nutr. 2014; 10(2):198–205. https://doi.org/10.1111/j.1740-

8709.2012.00407.x PMID: 22590991.

23. Haidar J, Omwega AM, Muroki NM, Ayana G. Daily versus weekly iron supplementation and prevention

of iron deficiency anaemia in lactating women. East Afr Med J. 2003; 80(1):11–6. https://doi.org/10.

4314/eamj.v80i1.8661 PMID: 12755236.

24. Tolkien Z, Stecher L, Mander AP, Pereira DIA, Powell JJ. Ferrous Sulfate Supplementation Causes Sig-

nificant Gastrointestinal Side-Effects in Adults: A Systematic Review and Meta-Analysis. PLOS ONE.

2015; 10(2):e0117383. https://doi.org/10.1371/journal.pone.0117383 PMID: 25700159

Community-based approach of iron and folic acid supplementation

PLOS ONE | https://doi.org/10.1371/journal.pone.0227351 January 10, 2020 15 / 17

25. Charoenlarp P, Dhanamitta S, Kaewvichit R, Silprasert A, Suwanaradd C, Na-Nakorn S, et al. A WHO

collaborative study on iron supplementation in Burma and in Thailand. Am J Clin Nutr. 1988;47.

26. Siekmans K, Roche M, Kung’u JK, Desrochers RE, De-Regil LM. Barriers and enablers for iron folic

acid (IFA) supplementation in pregnant women. Maternal & Child Nutrition. 0(0):e12532. https://doi.

org/10.1111/mcn.12532 PMID: 29271115

27. MCHIP. Community-based distribution for routine Iron/folic acid supplementation in pregnancy. Mater-

nal and Child Health Integrated Program, USAID, Nairobi: 2014.

28. Yip R. Iron supplementation: country level experiences and lessons learned. Journal of Nutrition. 2002;

132(859S-861S).

29. MICAH. Malawi final survey report. World Vision. 2006.

30. Lunet N, Rodrigues T, Correia S, Barros H. Adequacy of prenatal care as a major determinant of folic

acid, iron, and vitamin intake during pregnancy. Cadernos de Saude Publica. 2008; 24(5):1151–7.

https://doi.org/10.1590/s0102-311x2008000500022 PMID: 18461244.

31. Maina-Gathigi L, Omolo J, Wanzala P, Lindan C, Makokha A. Utilization of folic acid and iron supple-

mentation services by pregnant women attending an antenatal clinic at a regional referral hospital in

Kenya. Maternal Child Health Journal. 2013; 17(7):1236–42. Epub 2012/08/22. https://doi.org/10.1007/

s10995-012-1120-x PMID: 22907273.

32. Nisar Y, Michael JD, Ali MM. Factors associated with non-use of antenatal iron and folic acid supple-

ments among Pakistani women: a cross sectional household survey. BMC Pregnancy Childbirth. 2014;

14(1):305.

33. Harrison GG. Public health interventions to combat micronutrient deficiencies. Public Health Reviews.

2010; 32(1):256–66.

34. Kamau MW, Mirie W, Kimani S. Compliance with Iron and folic acid supplementation (IFAS) and associ-

ated factors among pregnant women: results from a cross-sectional study in Kiambu County, Kenya.

BMC public health. 2018; 18(1):580. https://doi.org/10.1186/s12889-018-5437-2 PMID: 29720135

35. Kamau M, Kimani S, Mirie W. Counselling on iron and folic acid supplementation (IFAS) is associated

with improved knowledge among pregnant women in a rural county of Kenya: a cross-sectional study

[version 1; referees: awaiting peer review]. AAS Open Research. 2018; 1(21). https://doi.org/10.12688/

aasopenres.12891.1

36. Demidenko E. Sample size and optimal design for logistic regression with binary interaction. Stat Med.

2008; 27(1):36–46. Epub 2007/07/20. https://doi.org/10.1002/sim.2980 PMID: 17634969.

37. Dinga LA. Factors associated with adherence to iron/folate supplementation among pregnant women

attending antenatal clinic at Thika District Hospital in Kiambu County, Kenya. Nairobi: University of Nai-

robi; 2013.

38. Arega Sadore A, Abebe Gebretsadik L, Aman Hussen M. Compliance with Iron-Folate Supplement and

Associated Factors among Antenatal Care Attendant Mothers in Misha District, South Ethiopia: Com-

munity Based Cross-Sectional Study. Journal of Environmental and Public Health. 2015; 2015:781973.

https://doi.org/10.1155/2015/781973 PMC4709613. PMID: 26839573

39. Kothari CR, Garg G. Research Methodology: Methods and Techniques:. Third Edition ed: New Age

International Limited Publishers; 2014.

40. Okube OT, Mirie W, Odhiambo E, Sabina W, Habtu M. Prevalence and factors associated with anemia

among pregnant women attending antenatal clinic in the second and third trimesters at Pumwani mater-

nity hospital, Kenya. Open journal of Obstetrics and Gynaecology. 2016; 6:16–27. doi: org/

10ojog.2016.61003

41. Mithra P, Unnikrishnan B, Rekha T, Nithin K, Mohan K, Kulkarni V, et al. Compliance with iron-folic acid

(IFA) therapy among pregnant women in an urban area of south India. African Health Sciences. 2013;

13(4):880–5. https://doi.org/10.4314/ahs.v13i4.3 PMC4056486. PMID: 24940307

42. Dairo MD, Lawoyin TO. Demographic factors determining compliance to iron supplementation in preg-

nancy in Oyo State, Nigeria. Niger J Med. 2006; 15.

43. Juma M, Oiye SO, Konyole SO. Predictors of optimum antenatal iron-folate supplementation in a low

resource rural set-up in Eastern Kenya. Journal of public health and Epidemiology. 2015; 7(11):337–45.

https://doi.org/10.5897/JPHE2015.0770

44. Gebreamlak B, Dadi AF, Atnafu A. High Adherence to Iron/Folic Acid Supplementation during Preg-

nancy Time among Antenatal and Postnatal Care Attendant Mothers in Governmental Health Centers

in Akaki Kality Sub City, Addis Ababa, Ethiopia: Hierarchical Negative Binomial Poisson Regression.

PLOS ONE. 2017; 12(1):e0169415. https://doi.org/10.1371/journal.pone.0169415 PMID: 28129344

45. Aguayo VM, Kone D, Bamba SI, Diallo B, Sidibe Y. Acceptability of multiple micronutrient supplements

by pregnant and lactating women in Mali. Public Health Nutrition. 2005; 8(1):33–7. https://doi.org/10.

1079/phn2005665 PMID: 15705243.

Community-based approach of iron and folic acid supplementation

PLOS ONE | https://doi.org/10.1371/journal.pone.0227351 January 10, 2020 16 / 17

46. Nisar YB, Alam A, Aurangzeb B, Dibley MJ. Perceptions of antenatal iron-folic acid supplements in

urban and rural Pakistan: a qualitative study. BMC Pregnancy and Childbirth. 2014; 14(1):1–12. https://

doi.org/10.1186/1471-2393-14-344 PMID: 25269515

Community-based approach of iron and folic acid supplementation

PLOS ONE | https://doi.org/10.1371/journal.pone.0227351 January 10, 2020 17 / 17